The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes text data from external sources like Reddit and Google News, constituting a potential surface for indirect prompt injection.
Ingestion points: External data enters the system through the search results fetched in scripts/scrapers.py from Reddit, Google News, and DuckDuckGo.
Boundary markers: The skill does not currently implement specific delimiters or instructions for the agent to ignore potential malicious prompts embedded within the retrieved brand mentions.
Capability inventory: The skill's capabilities are limited to performing sentiment analysis and generating reports; it does not currently perform sensitive actions such as arbitrary command execution, file system modifications, or exfiltration of local credentials.
Sanitization: Input content is filtered for sentiment-related keywords using regular expressions, but there is no comprehensive sanitization to prevent the AI agent from interpreting instructions contained within the brand mentions it monitors.

brand-monitoring